Gemini wears Prada

Inspiration

We wanted to make fashion technology feel less like a recommendation engine and more like an experience. We were inspired by the brutally honest, high-pressure world of The Devil Wears Prada and the idea of having an AI fashion critic that actually has a personality. Instead of simply asking AI, "Does this outfit look good?", we wanted it to feel like you're standing in front of a fashion editor waiting to hear whether your outfit is worthy of approval.

At the same time, we wanted to solve a problem almost everyone has experienced: having a closet full of clothes and somehow feeling like you have nothing to wear. That led us to Gemini Wears Prada an AI-powered fashion assistant that can critique your outfit, organize your wardrobe, and help you create new looks from the clothes you already own.

What it does

Gemini Wears Prada has two main features:

An AI Fashion Critic

Users can upload a photo of their outfit or take one directly through the app. Our AI analyzes the outfit and provides vocal, brutally honest fashion feedback through a Miranda Priestly-inspired character. Instead of receiving a generic rating, users get an entertaining fashion critique that comments on their outfit, styling, and overall look.

An AI Wardrobe

Users can photograph individual clothing items and add them to their digital wardrobe. The AI isolates the clothing and turns it into a sticker-style wardrobe item. Once their wardrobe is populated, users can:

Build outfits manually by mixing and matching their clothing
Generate a random outfit from their own wardrobe
Provide prompts such as dress code and season
Let a custom fashion algorithm fit you up

The goal is to make deciding what to wear easier, while making the process much more fun.

How we built it

We built Gemini Wears Prada as an interactive web application using React and Google's Gemini AI capabilities. Gemini handles the visual understanding needed to analyze outfit and clothing photos, identify fashion items, and generate outfit recommendations based on the user's wardrobe and prompts.

For the wardrobe feature, clothing photos are processed into individual visual assets that can be displayed as sticker-like pieces (via huggingface and rembg). These pieces can then be arranged into outfits or selected by the AI when generating a look. We also incorporated voice interaction via Eleven Labs to make the AI fashion critic feel more like a character rather than another chatbot. The combination of computer vision, generative AI, and voice makes the experience feel much more immersive.

Challenges we ran into

One of our biggest challenges was this is our second project, we scrapped our first project at around 7pm so we were under heavy time constraints but we persevered (I got 1.5 hours of sleep).